Blockwork wharves are the most common quay typology for existing port structures in the Mediterranean region. In this study, a comprehensive methodology for assessing the seismic vulnerability of such configurations is proposed, and an application to an important Italian maritime hub is presented. The geotechnical system made up by the foundation system, the backfill soil and the blockwork wall is modelled using FLAC 2D, an advanced fully nonlinear, finite difference software, based on an explicit Lagrangian calculation scheme. Extensive time histories analyses are performed for different earthquake intensity levels, corresponding to several return periods. The damage to the facility is assessed according to the damage criterion expressed by the International Navigation Association (PIANC), which is based on selected Engineering Demand Parameters (EDPs). The EDPs’ sensitivity to each of the model input parameters, to the earthquake level, and to the ground motion is then analyzed by means of Tornado (swing) diagrams and First Order Second Moment (FOSM) analysis. An Artificial Neural Network (ANN) that considers as input data the geotechnical properties, the structural configuration and selected Intensity Measures (IMs), is then implemented to evaluate the seismically induced damage. The ANN prediction is validated by partitioning the inner weight matrices, and such sensitivity is compared to the statistical results obtained using the Tornado and FOSM techniques. Based on the ANN, design and assessment charts can be systematically derived. Finally, a novel approach to originate fragility curves is illustrated and results are presented for the selected case study.